High-Risk Care Management Impact on Medicaid ACO Utilization and Spending

ABSTRACT

Objectives: States are experimenting with accountable care organization (ACO) contracts to slow Medicaid spending growth. There is limited information on how ACOs have impacted expenditures in Medicaid, which includes a relatively heterogenous population with less spending compared with Medicare. This study aimed to evaluate the impact of high-risk care management on spending and utilization within Massachusetts’ largest Medicaid ACO.

Study Design: This observational study analyzed Medicaid claims data from Massachusetts’ largest Medicaid ACO utilizing staggered program entry from 2016 to 2021 (n = 158,441 total beneficiaries). It included adults aged 18 to 64 years with multiple chronic conditions and used a claims-based algorithm for participant selection. We examined spending and clinical event rates of those participating and not yet participating in a high-risk care management program. Between 2016 and 2021, 2479 beneficiaries were identified as high risk and entered the program.

Methods: The study utilized a difference-in-differences approach with linear regression models to assess the impact of care management. The analysis accounted for time-stable and time-changing covariates, including comorbidity levels and age.

Results: Participation in the program for 7 or more months was associated with a $243 reduction in monthly spending compared with similar beneficiaries who had not yet started the program (95% CI, –$479 to –$6). There also were comparable reductions in emergency department visit and hospitalization rates.

Conclusions: These early ACO data findings suggest that upstream care management of high-risk individuals may represent a viable approach for slowing Medicaid spending growth.

Am J Manag Care. 2025;31(7):In Press

Takeaway Points

High-risk care management reduced health care costs and utilization within Massachusetts’ largest Medicaid accountable care organization (ACO).

• High-risk care management significantly lowered total medical spending, emergency department visits, and hospital admissions.
• Notable cost and utilization reductions emerged after 6 months.
• This strategy can serve as a model for other states, especially in an evolving health care landscape post COVID-19.
• The study’s findings support Medicaid ACOs and targeted care management as a viable strategy to control spending growth without compromising care quality.
• These insights may aid states seeking sustainable health care reforms, especially under budget constraints and shifting Medicaid funding.

Health care costs have grown over the past decade, with state Medicaid expenditures consuming larger shares of state budgets despite expanding federal contributions.1 According to CMS, Medicaid expenses as a percentage of state budgets increased from 12.5% to 28.3% between 1990 and 2020.2 Upward trends in costs can be attributed to Medicaid enrollment growth, rising health service costs with fee-for-service payment models, and the COVID-19 pandemic and subsequent price increases secondary to rapid inflation.3-5 Additional contemporaneous changes include Medicaid disenrollment and reduced reimbursement with the end of the public health emergency (PHE), resulting in reduced state Medicaid funds.6 As a result, states are experimenting with alternative payment models to improve value.

Efforts to bend the Medicaid cost curve at times have been at odds with providing high-quality care. States have attempted to reduce spending by reducing the number of enrollees (eg, by restricting eligibility criteria or increasing the frequency of eligibility checks), limiting the number of prescription medications covered per beneficiary, and reducing provider payments.4 These efforts have had variable impact on total spending and little impact on per capita spending while reducing access to care.7 These unsuccessful attempts led to a proliferation of alternative Medicaid payment models, including bundled payments and episodes of care, global capitated payments, managed care, and more recently, accountable care organizations (ACOs) with shared savings and risk.8

Little is known about the efficacy of nascent Medicaid ACOs, and there are reasons to be skeptical that they could reduce spending growth. Compared with Medicare, Medicaid programs comprise a heterogeneous mix of children and healthy adults, who have lower spending at baseline, and disabled and older people, who have a higher spending pattern. Further, this population is poorer and experiences more barriers that might require social and structural investments outside Medicaid to prevent downstream medical spending. Moreover, although a handful of studies have evaluated the performance of individual ACOs across the US, there are limited data on the impacts of Medicaid ACOs on spending growth and none on the impact of specific strategies used by ACOs on downstream individual spending.9-11

A common quality-enhancing, cost-reducing strategy that insurers and health systems have used is high-risk care management. Within the health system that is the focus of this study, participation in a care management program (CMP) reduced spending and utilization in its Medicare Pioneer ACO in a dose-dependent manner.12

This evaluation examines the impact of a health system–based CMP for high-risk individuals enrolled in a Medicaid ACO. High-risk care management served as the primary mechanism for reducing spending growth within the ACO.

METHODS

Study Data and Methods

We used Massachusetts Medicaid claims data for 2016 through 2021 provided by MassHealth (the Massachusetts Medicaid program) for the Mass General Brigham (MGB) Medicaid ACO. The MGB multisite health system in the eastern half of Massachusetts has participated in multiple ACOs and used a CMP to support them. Our ACO study population consisted of MassHealth members with an MGB-assigned primary care physician who were aligned with either the Massachusetts Medicaid ACO Pilot (August 2016-February 2018) and/or the Massachusetts Medicaid ACO Program (March 2018-December 2021).

For our study period, an MGB claims-based algorithm, run once a year, identified individuals with multiple chronic conditions, high levels of health care utilization, and predicted high utilization in the coming year and referred them to the high-risk CMP.13 This algorithm, which has been described in previous publications, uses a machine learning approach that incorporates both claims data and clinical information, including variables such as chronic conditions, health care utilization, and history of missed appointments, to identify patients who are most likely to benefit from care management (using the Johns Hopkins ACG System).13-18 Primary care physicians reviewed the list of identified high-risk patients generated by the algorithm and added or removed patients based on clinical judgment. Care managers, such as registered nurses, licensed independent clinical social workers, and community health workers, were assigned to patients based on medical vs behavioral complexity. They enrolled patients, conducted baseline assessments, developed care plans, and coordinated team-based care management. Patients could be removed from the program if they did not agree to participate, graduated, were lost to follow-up, or died. The program is embedded in primary care and consists of in-person, virtual, and telephonic contact with patients.

This study included adults aged 18 to 64 years who were enrolled in the MGB MassHealth ACO and had a primary care physician within the MGB network. All members who joined the CMP during the observation period were included, regardless of whether they participated in the pilot or full ACO. Participation start date was defined using electronic health records as the date the care manager determined the member was eligible for the program. This program was specific to MGB’s MassHealth ACO patients and did not include all Medicaid members.

To assess the association between CMP exposure time and outcomes, we calculated the number of months each beneficiary received CMP care. We included only those who had at least 1 month of Medicaid ACO enrollment before being in the CMP and at least 1 month of follow-up after the month of CMP entry. Our study population included individuals in the Medicaid ACO who joined the CMP from January 2018 to November 2021. For these individuals, we included all available months of claims data at any point while enrolled in the ACO from August 2016 through December 2021. We also conducted a sensitivity analysis restricting the population to those with 12 months or more of continuous enrollment in the ACO both before and after CMP entry (ie, 24 continuous months of enrollment centered around the CMP entry month).

Time-Stable and Time-Changing Covariates

Because of concerns about unmeasured time-stable confounding, such as by the level of education, we included individual-level fixed effects in our analyses. This approach accounts for differences across comparison groups in unobserved characteristics that did not change over time.

Annually updated time-varying covariates of interest included changes in comorbidity levels and age. To adjust for comorbidities, we used the HHS Hierarchical Condition Categories (HCCs).19 We calculated risk scores using inpatient and outpatient diagnostic codes at ACO entry and recalculated annually for each year of ACO enrollment.16

Outcomes

The primary outcomes were total monthly medical spending, defined as the sum of inpatient and outpatient spending (excluding pharmacy spending), monthly emergency department (ED) visits, and monthly inpatient admissions. We used medical spending claims to calculate total monthly medical spending for each individual, standardizing costs to 2021 US$ using the Consumer Price Index. We counted ED visits and inpatient admissions per month using claims data. We allowed for a maximum of 1 type of each visit per date, meaning if there were 2 ED visits on the same date (eg, a patient is discharged from the ED and then returns on the same date), we counted that date as having a single ED visit. Outcomes were compared with the mean spending and utilization of all ACO months prior to CMP assessment.

Analyses

We conducted difference-in-differences analyses with linear regression models with individual-level fixed effects to compare outcomes of those with early vs later CMP participation start. This approach used a parallel trend assumption. This same approach was used to assess the mean CMP effect among Medicare Pioneer ACO members in past publications.12,18

Staggered CMP entry dates effectively created natural control groups whose members similarly qualified and received care in the same health system but who had not yet entered the program at a given point in time. CMP enrollees who started in each month were compared with those who started in subsequent months. Our predictor variable was the time-changing indicator for pre– vs post CMP participation. The month of CMP start was excluded because it often included days prior to CMP participation and with CMP participation.

We assessed the pre-CMP visit rate and spending trends for those entering the program early vs later in time both visually and using a logistic regression model with an interaction term for calendar time and timing of entry. The outcome trends were comparable.

For all outcomes, we used linear models with a fixed subject effect and robust SEs. We defined exposure to CMP in 2 ways. In model 1, we employed a binary pre– and post participation start definition. In model 2, we created a 3-level indicator of time in CMP: pre-CMP vs 1 to 6 months in CMP vs 7 or more months in CMP.

We adjusted for year, month, and 8 time-varying HCC indicators hypothesized to affect spending and used in prior studies: diabetes with chronic complications, diabetes without chronic complications, major depressive disorder and/or bipolar affective disorder, seizure disorders and convulsions, congestive heart failure, chronic obstructive pulmonary disease, asthma, and completed pregnancy with no or minor complications.18

Sensitivity Analyses

We conducted several sensitivity analyses. First, we examined the same outcomes in members who were continuously enrolled in the ACO 12 months before and after CMP entry. Second, for a pre–COVID-19 sensitivity analysis (January 2018-January 2020), we included only those with continuous ACO enrollment for 12 months prior to CMP start. In the third sensitivity analysis, we employed inverse probability weighting to address possible nonrandom disenrollment from the ACO and CMP. We examined the probability of staying enrolled through the end of the study period (December 2021) and gave more weight to those patients who disenrolled from the ACO.

RESULTS

Between August 2016 and December 2021, the Medicaid ACO enrolled 2479 adults who were identified as having a high risk for future medical spending and were thus eligible for the CMP. The Medicaid ACO provided care for 155,962 adults who did not participate in CMP (Table 1). The majority of high-risk patients were female, and the mean age was 45 years.

Spending

After adjustment, monthly spending was $243 less (95% CI, –$479 to –$6) for high-risk patients receiving CMP care for at least 7 months compared with similar patients who had yet to receive CMP care (Table 2). This represents a 16% reduction in spending compared with the mean spending of $1522 in all ACO months prior to CMP assessment. There were no statistically significant differences in spending overall or during months 1 through 6 of program enrollment compared with spending for those not yet in the program.

In a sensitivity analysis of those with continuous enrollment in the Medicaid ACO for 12 months prior to CMP entry and 12 months following CMP entry, we found an overall reduction in mean monthly spending of $220 (95% CI, –$410 to –$30) and a reduction of $330 a month (95% CI, –$570 to –$91) among patients enrolled for 7 or more months compared with spending for patients not yet in the program. This continuous enrollment sensitivity analysis accounted for any differences due to mortality. In analyses accounting for potential differential disenrollment prior to December 2021, the CIs spanned zero.

ED Visits

Among high-risk patients, ED visit rates were lower by 0.039 visits per month (95% CI, –0.050 to –0.028) for those receiving CMP care vs those who had not yet received it (Table 2). Reductions were significant during months 1 to 6 with 0.033 fewer visits per month (95% CI, –0.044 to –0.022) and at 7 or more months with 0.052 fewer visits per month (95% CI, –0.065 to –0.038) compared with those who had not yet entered the program. Overall, there was a 30% reduction in mean ED visits per month for CMP-enrolled members compared with the mean for the ACO months prior to CMP enrollment.

Sensitivity analyses including only those with continuous enrollment in the Medicaid ACO for 12 months prior to CMP enrollment and 12 months following CMP entry showed a similar reduction in ED visits overall and during months 1 to 6 post enrollment and 7 or more months post enrollment compared with those who had not yet entered the program. Analyses that accounted for potential differential attrition also had consistent results.

Inpatient Admissions

For those receiving CMP care, monthly inpatient admissions were lower by 0.012 visits per month (95% CI, –0.017 to –0.007) overall compared with those who had not yet started the CMP (Table 2). Inpatient admission rates were 0.016 per month lower (95% CI, –0.023 to –0.010) after 7 or more months of CMP enrollment compared with those who had not yet started the CMP. Overall, this represented a 30% reduction in mean hospital admissions per month for CMP-enrolled members compared with the mean for the ACO months prior to CMP enrollment.

Sensitivity analyses including only those patients with continuous enrollment in the Medicaid ACO for 12 months prior to CMP enrollment and 12 months following CMP entry and analyses accounting for potential differential disenrollment prior to December 2021 showed similar results.

DISCUSSION

To our knowledge, this is the first evaluation of a Medicaid ACO program on spending growth that targets the specific program designed to alter spending patterns in a high-risk group participating in care management.

Unlike traditional fee-for-service payment models, ACOs incentivize both cost containment and improved health outcomes via risk sharing because the payer and the provider share cost savings and cost overruns relative to benchmarks.20-22 In 2020, Medicare Shared Savings Program ACOs achieved $1.861 billion in net savings compared with their benchmarks, recording their fourth consecutive year of net savings.21 Meanwhile, despite mixed results for commercial ACOs, major insurers have increased their involvement in them substantially.23,24

These trends catalyzed many states to pilot ACO payment models in Medicaid. CMS authorized states to offer innovative Medicaid ACO programs through Section 1115 waivers. In 2016, Massachusetts became 1 of the first 6 states to develop their ACO programs using federal funding.1 By December 2018, the state’s Medicaid ACOs served 65% of beneficiaries, more than 1.2 million people.25 At the same time, other states were creating and quickly ending ACO experiments due to a lack of quick returns on investment and poorly coordinated rollouts.26,27

Current studies of care management impact within Medicaid ACOs have limited follow-up time and evaluation of specific services.28,29 In the Camden Coalition ACO, which covered 4 major New Jersey health systems and a high-risk community case management program, the annual savings rate ranged between 0.4% and 3.2% using an outlier ceiling of $100,000 and a variable trend factor.11 In a randomized controlled trial of the Camden Coalition care management program, there was no change in hospital readmissions but a 15% increase in postdischarge ambulatory visits.30 Overall, these evaluations showed lower-than-expected savings, illustrating the importance of capturing areas where ACOs have the biggest impact in the highest-risk, highest-utilizing population with sufficient follow-up time.

Our major findings support state participation in a Medicaid ACO and the specific strategy of using CMPs for high-risk, high-utilizer groups to control spending growth. Our findings are similar to previous Medicare ACO care management evaluations, with a time-in-program–dependent reduction in utilization and spending.12 First, we found a reduction in ED utilization within the first 6 months and inpatient admissions after 6 months in the program. This was followed by a reduction in spending seen after at least 6 months in the CMP.

Second, our findings are consistent with the importance of policies that disincentivize population turnover in ACOs. Consistent care improves patient-provider continuity and longitudinal chronic care and encourages health systems to invest in outcomes that have longer payoff times.31 With higher levels of turnover and shorter tenures in ACOs, the incentives for reducing individual downstream spending are muted, whereas the incentives for skimping on care could be more prominent. Sensitivity analysis findings are consistent with the trend toward improved spending and utilization, although statistical significance is limited due to small sample size and limited follow-up time.

Placing our findings in a larger context, with the end of the national PHE in May 2023, the end of continuous Medicaid coverage, and the passage of the Consolidated Appropriations Act of 2023, states experienced further pressure to control Medicaid spending. CMS slowly wound down the 6.2% increase in the regular federal matching rate (ie, Federal Medical Assistance Percentage [FMAP]) by December 31, 2023.6 The FMAP dictates the federal portion of Medicaid reimbursement for health care services, and the increase in funding was critical to support the spike in Medicaid enrollment throughout the COVID-19 pandemic.32 With the discontinuation of the FMAP increase, mass disenrollment from Medicaid, estimated to be 25 million nationwide, resulted in an absolute reduction in dollars that states receive per beneficiary as well as a relative reduction in federal dollars per beneficiary.33,34 With fewer beneficiaries, the mean fixed cost per beneficiary could increase, which could exacerbate the cost pressures created by the decreased federal contribution per beneficiary.

Although states may have a cushion due to leftover pandemic dollars, we have yet to see the extent of the pandemic’s toll on the economy. States will be looking for alternative ways to cut spending while maintaining patient quality. With the Trump administration’s support for states to submit Section 1115 waivers and focus on preventive care, states may consider piloting a Medicaid ACO program.35 Our findings suggest that targeted care management is an important mechanism for reducing spending in such an ACO.

Limitations

There were several limitations to this analysis. First, this is a single health system in a single state, so generalizability to other systems or states could be limited. However, the MGB Medicaid ACO represents the largest ACO in Massachusetts, with approximately 158,000 beneficiaries.

Second, there may have been nonrandom CMP participation initiation over time. The high-risk algorithm selects beneficiaries with multiple chronic conditions and high utilization. Due to program capacity limitations, the algorithm identifies more patients than could participate at a given time. It is possible there was selection bias in who joined earlier vs later, potentially biasing outcomes if the earliest participating patients had more to gain than subsequently added patients. Alternatively, participants starting the CMP later, when there were greater capacity constraints, may have had higher needs and therefore more to gain from the CMP. However, we found no association between predicted spending levels and entry timing. Lastly, although participation in a Medicaid ACO is often invisible to enrollees, participation in care management was voluntary throughout the study and could result in bias.

Third, in the analysis, we assumed independence of all time-stable covariates. For any interaction there may have been between a beneficiary trait and the intervention, we assumed a constant proportion of beneficiaries with the trait between early and late enrollees. This could introduce some unmeasured bias.

Fourth, our analysis included all Medicaid medical spending, although it did not include CMP costs. The ACO did not collect information about incremental costs associated with program entry. Because of the integrated nature of the program, separating it from routine care costs was beyond this study’s scope.

Lastly, the COVID-19 pandemic altered both Medicaid enrollment patterns and medical use. We did not have sufficient statistical power to isolate the pandemic-era effects from 2020 to 2022; in our prepandemic sensitivity analysis from 2018 to 2019, we saw similar trends in decreased utilization and spending.

CONCLUSIONS

High-risk care management reduced spending, ED visit rates, and hospitalization rates among high-risk patients enrolled in a Medicaid ACO. As expected, the benefits of care management accrued over time, with greater reductions in spending and utilization after 6 months of care management—showing that altering downstream care needs required time. Future research should focus on how care management strategies adapt in the aftermath of the PHE to ensure sustained cost savings and improved patient outcomes. n

Acknowledgments

Kathryn Corelli, MD, and Erin Duralde, MD, contributed equally to this work and are listed as co–first authors.

Author Affiliations: Population Health Management, Mass General Brigham (KC, CV, MMV, LJ, GSM), Somerville, MA; Division of General Internal Medicine, Massachusetts General Hospital (KC, KHS), Boston, MA; Department of Healthcare Policy, Harvard Medical School (KC, ED, CV, JH), Boston, MA; Division of Women’s Health (ED) and Division of Nephrology (MLM), Brigham and Women’s Hospital, Boston, MA; Geisel School of Medicine, Dartmouth College (DM), Hanover, NH; Mongan Institute, Massachusetts General Hospital (MP, NMB, NA, VF, CV, JH), Boston, MA; Harvard University (NA), Cambridge, MA; McLean Hospital (NMB), Belmont, MA.

Source of Funding: HHS grant funding (P01AG032952)

Author Disclosures: Dr Fung has received grants from the National Institutes of Health and has a grant pending with the Agency for Healthcare Research and Quality. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (KC, ED, NMB, MLM, GSM, JH); acquisition of data (KC, ED, CV, LJ, GSM, JH); analysis and interpretation of data (KC, ED, DM, MP, NMB, NA, VF, JH); drafting of the manuscript (KC, ED, DM, MP, NA, CV, MLM); critical revision of the manuscript for important intellectual content (KC, ED, DM, NMB, NA, VF, CV, KHS, MMV, MLM, GSM, JH); provision of patients or study materials (CV, KHS); obtaining funding (MMV, LJ); administrative, technical, or logistic support (MP, VF, CV, KHS, GSM); and supervision (LJ, MMV, GSM, JH).

Address Correspondence to: Kathryn Corelli, MD, Mass General Brigham, 399 Revolution Dr, Somerville, MA 02145. Email: kmcorelli@mgb.org.

REFERENCES

1. Matulis R, Lloyd J. The history, evolution, and future of Medicaid accountable care organizations. Center for Health Care Strategies brief. February 2018. Accessed October 6, 2022. https://www.chcs.org/media/ACO-Policy-Paper_022718.pdf

2. Medicaid expenses as a percent of state budgets. State Health Access Data Assistance Center. Accessed October 7, 2022. https://statehealthcompare.shadac.org/map/103/medicaid-expenses-
as-a-percent-of-state-budgets#a/32/140

3. Williams E. Medicaid enrollment & spending growth: FY 2022 & 2023. KFF. October 25, 2022. Accessed January 5, 2023. https://www.kff.org/medicaid/issue-brief/medicaid-enrollment-spending-growth-fy-2022-2023/

4. Understanding Medicaid: a primer for state legislators. National Conference of State Legislatures. January 5, 2021. Accessed October 7, 2022. https://web.archive.org/web/20220702044209/https://www.ncsl.org/research/health/understanding-medicaid-a-primer-for-state-legislators.aspx

5. NHE fact sheet. CMS. Updated August 12, 2022. Accessed October 7, 2022. https://www.cms.gov/Research-
Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/NHE-Fact-Sheet

6. Cubanski J, Kates J, Tolbert J, Guth M, Pollitz K, Freed M. What happens when COVID-19 emergency declarations end? implications for coverage, costs, and access. KFF. January 31, 2023. Accessed February 8, 2023. https://www.kff.org/coronavirus-covid-19/issue-brief/what-happens-when-covid-19-emergency-declarations-end-implications-for-coverage-costs-and-access/

7. Tu T, Muhlestein D, Kocot SL, White R. Origins and future of accountable care organizations. Brookings Institution. May 2015. Accessed October 7, 2022. https://www.brookings.edu/wp-content/uploads/2016/06/Impact-of-Accountable-CareOrigins-052015.pdf

8. Bazell, Alston M, Pelizzari P, Sweatman B. What are bundled payments and how can they be used by healthcare organizations? Milliman Inc. March 27, 2023. Accessed June 20, 2023. https://www.milliman.com/en/insight/what-are-bundled-payments-and-how-can-they-be-used-by-healthcare-organizations

9. Vickery KD, Shippee ND, Guzman-Corrales LM, et al. Changes in quality of life among enrollees in Hennepin Health: a Medicaid expansion ACO. Med Care Res Rev. 2020;77(1):60-73. doi:10.1177/1077558718769457

10. Brown M, Ofili EO, Okirie D, et al. Morehouse Choice Accountable Care Organization and Education System (MCACO-ES): integrated model delivering equitable quality care. Int J Environ Res Public Health. 2019;16(17):3084. doi:10.3390/ijerph16173084

11. Truchil A, Dravid N, Singer S, Martinez Z, Kuruna T, Waulters S. Lessons from the Camden Coalition of Healthcare Providers’ first Medicaid Shared Savings performance evaluation. Popul Health Manag. 2018;21(4):278-284. doi:10.1089/pop.2017.0164

12. Hsu J, Price M, Vogeli C, et al. Bending the spending curve by altering care delivery patterns: the role of care management within a Pioneer ACO. Health Aff (Millwood). 2017;36(5):876-884. doi:10.1377/hlthaff.2016.0922

13. Vogeli C, Spirt J, Brand R, et al. Implementing a hybrid approach to select patients for care management: variations across practices. Am J Manag Care. 2016;22(5):358-365.

14. Haime V, Hong C, Mandel L, et al. Clinician considerations when selecting high-risk patients for care management. Am J Manag Care. 2015;21(10):e576-e582.

15. Hsu J, Price M, Spirt J, et al. Patient population loss at a large Pioneer accountable care organization and implications for refining the program. Health Aff (Millwood). 2016;35(3):422-430. doi:10.1377/hlthaff.2015.0805

16. Yun H, Kilgore ML, Curtis JR, et al. Identifying types of nursing facility stays using Medicare claims data: an algorithm and validation. Health Serv Outcomes Res Methodol. 2010;10:100-110. doi:10.1007/s10742-010-0060-4

17. Institute of Medicine. The Healthcare Imperative: Lowering Costs and Improving Outcomes: Workshop Series Summary. The National Academies Press; 2010. doi:10.17226/12750

18. Benson NM, Price M, Weiss M, et al. Tacking upwind: reducing spending among high-risk commercially insured patients. 2023;29(5):220-226. doi:10.37765/ajmc.2023.89355

19. Kautter J, Pope GC, Ingber M, et al. The HHS-HCC risk adjustment model for individual and small group markets under the Affordable Care Act. Medicare Medicaid Res Rev. 2014;4(3):E1-E46. doi:10.5600/mmrr.004.03.a03

20. Levinson D. Medicare Shared Savings Program Accountable Care Organizations Have Shown Potential for Reducing Spending and Improving Quality. HHS Office of Inspector General; August 2017. Accessed October 7, 2022. http://resource.nlm.nih.gov/101737819

21. Highlights of the 2020 Medicare ACO program results. National Association of ACOs. Accessed October 7, 2022. https://www.naacos.com/highlights-of-the-2020-medicare-aco-program-results

22. Hsu J, Vogeli C, Price M, et al. Substantial physician turnover and beneficiary ‘churn’ in a large Medicare Pioneer ACO. Health Aff (Millwood). 2017;36(4):640-648. doi:10.1377/hlthaff.2016.1107

23. Peiris D, Phipps-Taylor MC, Stachowski CA, et al. ACOs holding commercial contracts are larger and more efficient than noncommercial ACOs. Health Aff (Millwood). 2016;35(10):1849-1856. doi:10.1377/hlthaff.2016.0387

24. Our take: commercial ACOs top 3,000 agreements, 30 million members. Darwin Research Group. August 19, 2019. Accessed October 7, 2022. https://www.darwinresearch.com/commercial-acos-top-3000-agreements-
30-million-members/

25. Medicaid Managed Care Enrollment And Program Characteristics, 2018. CMS Division of Managed Care Operations; summer 2021. Accessed October 7, 2022. https://www.medicaid.gov/medicaid/managed-care/downloads/2018-medicaid-managed-care-enrollment-report-updated.pdf

26. DeLia D, Yedidia MJ. The policy and practice legacy of the New Jersey Medicaid ACO demonstration project. J Ambul Care Manage. 2020;43(1):2-10. doi:10.1097/JAC.0000000000000308

27. States that reported accountable care organizations in place. KFF. Accessed July 12, 2022. https://www.kff.org/medicaid/state-indicator/states-that-reported-accountable-care-organizations-in-place/

28. Rutledge RI, Romaire MA, Hersey CL, Parish WJ, Kissam SM, Lloyd JT. Medicaid accountable care organizations in four states: implementation and early impacts. Milbank Q. 2019;97(2):583-619. doi:10.1111/1468-0009.12386

29. McConnell KJ, Renfro S, Chan BK, et al. Early performance in Medicaid accountable care organizations: a comparison of Oregon and Colorado. JAMA Intern Med. 2017;177(4):538-545. doi:10.1001/jamainternmed.2016.9098

30. Finkelstein A, Cantor JC, Gubb J, et al. The Camden Coalition Care Management Program improved intermediate care coordination: a randomized controlled trial. Health Aff (Millwood). 2024;43(1):131-139. doi:10.1377/hlthaff.2023.01151

31. Benson NM, Price M, Vogeli C, et al. Population turnover and leakage in commercial ACOs. Am J Manag Care. 2023;29(4):e104-e110. doi:10.37765/ajmc.2023.89350

32. Guth M, Rudowitz R, Garfield R. Federal Medicaid outlays during the COVID-19 pandemic. KFF. April 27, 2021. Accessed March 14, 2023. https://www.kff.org/coronavirus-covid-19/issue-brief/federal-medicaid-outlays-during-the-covid-19-pandemic/

33. Tolbert J, Ammula M. 10 things to know about the unwinding of the Medicaid continuous enrollment requirement. KFF. October 21, 2022. Accessed November 8, 2022. https://web.archive.org/web/
20221108010922/https://www.kff.org/medicaid/issue-brief/10-things-to-know-about-the-unwinding-of-the-medicaid-continuous-enrollment-requirement/

34. As Medicaid unwinding concludes in most states, KFF finds 25 million lost Medicaid coverage but enrollment is 10 million higher than pre-pandemic levels. News release. KFF. September 18, 2024. Accessed June 1, 2025. https://www.kff.org/medicaid/press-release/as-medicaid-unwinding-concludes-in-most-states-kff-finds-25-million-lost-medicaid-coverage-but-enrollment-is-10-million-higher-than-pre-pandemic-levels/

35. Hut N. Where value-based payment is headed during the second Trump administration. HFMA. March 3, 2025. Accessed June 1, 2025. https://www.hfma.org/where-value-based-payment-is-headed-during-the-second-trump-administration/